Backward compatibility for checkpoint loading (Lightning-AI#1132)

* check if hparams_type exists in checkpoint dictionary for backward compatibility

* concisely maintain backward compatibility for hparams type

* Bug fix in checkpoint loading (Lightning-AI#1132)

Author: amoudgl

CHANGELOG.md

@@ -29,6 +29,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 ### Fixed
 
 - Fixed bug related to type cheking of `ReduceLROnPlateau` lr schedulers([#1114](https://github.com/PyTorchLightning/pytorch-lightning/issues/1114))
+- Fixed a bug to ensure lightning checkpoints to be backward compatible ([#1132](https://github.com/PyTorchLightning/pytorch-lightning/pull/1132))
 
 ## [0.7.1] - 2020-03-07
 

pytorch_lightning/core/lightning.py

@@ -1396,7 +1396,7 @@ def _load_model_state(cls, checkpoint: Dict[str, Any]) -> 'LightningModule':
 
         if cls_takes_hparams:
             if ckpt_hparams is not None:
-                is_namespace = checkpoint.get('hparams_type') == 'namespace'
+                is_namespace = checkpoint.get('hparams_type', 'namespace') == 'namespace'
                 hparams = Namespace(**ckpt_hparams) if is_namespace else ckpt_hparams
             else:
                 warnings.warn(